HNHacker News
TopNewBestAskShowJobs

rahimnathwani

16,488 karma · joined March 9, 2012

I've lived in San Francisco since 2019.

2010-2019 I lived in China.

Before that I lived in London, where I was born.

My personal web site: https://encona.com/

Random tools: https://tools.encona.com/

Other random stuff about me:

- English is my native language, but I also speak (to varying degrees of fluency) Mandarin, Gujarati and Spanish.

- My career so far has been a mix of startups/cofounding, big tech (I was a PM at Amazon and at Google), and exec roles.

- I'm a qualified accountant (CIMA), have two undergrad degrees (one in CS) and an MBA (sorry!).

- I got married in 2012 and our son was born in 2016 (I spend a lot of time thinking about how best to raise him to be a capable and kind adult)

Email: rahim AT encona DOT com

LinkedIn: http://www.linkedin.com/in/rahimnathwani

X: https://x.com/rahimNathwani/

submissionscomments
rahimnathwani··on Show HN: Science for Kids
https://news.ycombinator.com/item?id=38092081
rahimnathwani··on Show HN: Science for Kids
I'm close to satisfied with the article text.

I'm still experimenting with the best ways to generate the images. I'm very happy with some of them, but unhappy with others. I need to figure out how to create an effective review loop. As you say, I'm not familiar enough with all of the subjects. Like I can't tell whether a diagram of a hearts has all the tubes in the right place.

I think AI is ideally suited to something like this, and I should soon be able to get to a point where I can generate an article, without human review, an article that's as good as that in a median kids' print magazine.

If I reach that point, the next goal would be to match or exceed the quality of the median article in Oyla (https://oyla.us/) or Aquila (https://aquila.co.uk/).

rahimnathwani··on Show HN: Science for Kids
There's a print-friendly stylesheet, so you can easily print each article. You could probably automated this using the RSS feed.

But I'm not sure I'd want a printable book. The articles are each standalone so you'd need to figure out some logical sequence. Plus once you print it you can't make improvements to existing articles.

rahimnathwani··on Silicon Valley sees AI as the solution – for everyone else
In the US, ~80% of the prime working age population is employed:

https://fred.stlouisfed.org/series/LNS12300060

For every one person who is classified as unemployed, there are one or more others who are classified unable to work for some reason.

rahimnathwani··on Twenty Years of Pandoc
I just noticed when I right-click in Outlook for Mac there's a 'Paste markdown' option. I don't know how long that's been there...
rahimnathwani··on Pi's Minimalism Is Its Advantage
It was introduced in late 2025:

https://github.com/earendil-works/pi/issues/92

rahimnathwani··on Pi's Minimalism Is Its Advantage
Mario recommends you don't do that.

You could try ohmypi but it sort of misses the point.

Why not just install nicopreme's pi-subagents and pi-web-access and then install whatever else you need when you find it's missing?

rahimnathwani··on Pi's Minimalism Is Its Advantage
Pi has auto compaction enabled by default:

https://pi.dev/docs/latest/compaction

rahimnathwani··on That time when I failed the Microsoft interview
From your descriptions they don't seem relevant to this discussion.
rahimnathwani··on That time when I failed the Microsoft interview
Your last statement is closer to the real situation, but you make it sound like it's certain that you will be called upon to defend the legality of your actions.

But I'm not aware of any evidence showing that is true.

And, even if it were true in the past, it seems that the DOJ believes that was based on a misinterpretation of the law. So it is less likely be true since June 2026: https://www.justice.gov/olc/media/1444871/dl

rahimnathwani··on That time when I failed the Microsoft interview
Is there a specific page on his blog that answers my question?

His suggestions for improving public transport are not relevant to this discussion.

rahimnathwani··on That time when I failed the Microsoft interview
"illegal unless it has passed through a battery of compliance paperwork"

"This is literally a Title VII violation."

Griggs v. Duke Power Co. held that "The [Civil Rights] Act [of 1964] does not preclude the use of testing or measuring procedures, but it does proscribe giving them controlling force unless they are demonstrably a reasonable measure of job performance."

Why do you believe that prior documentation is required, and that the absence of such documentation would make the testing illegal?

rahimnathwani··on Kelly Criterion Simulator
You are right! It's stupid to have a target.

The target wasn't there in the original version... I'm embarrassed to say I asked Claude to do a bunch of general 'make this better' on a few of the items on that site, and didn't check the output closely :(

rahimnathwani··on Twenty Years of Pandoc
Pandoc is great. I use it multiple times per day to move content between Outlook emails and a coding harness:

https://gist.github.com/rahimnathwani/210b1f9cb6ce731a304322...

  cat email-to-xyz.md | md2clip

  clip2md > email-from-xyz.md
rahimnathwani··on Kelly Criterion Simulator
I made a 'game' to help me son get some intuition for bet sizing: https://tools.encona.com/investor
rahimnathwani··on Dear Alpha School: I Hope You're Right
Whether they can scalably deliver these results is a next-level question.

I'd like an answer to a more basic one:

Of Alpha's current students, what proportion hit 2x or more of their Fall-projected growth on every Spring MAP test they took? (Across at least math and ELA.)

Joe confirmed that this is the specific commitment Alpha makes to parents, but no one from Alpha has, as far as I can tell, been willing to state how well they're meeting that commitment.

rahimnathwani··on On the non-use of AI in my writing process
He doesn't want to use a model trained on someone else's text. So he would need to train a model from scratch. I don't know whether his own texts would provide enough data.
rahimnathwani··on Show HN: Ski – Voice Coding for Claude Code, Codex and More – On-Device – Free
In the hero image the message on the left says 'Run the tests and tell me what broke'.
rahimnathwani··on The session you cannot take with you
I see your point. I was thinking mainly about #3 and how nothing, not even pi, can solve that when the reasoning traces are encrypted by the model provider.
rahimnathwani··on The session you cannot take with you
"I’ve essentially built a software factory using Pi and a custom terminal multiplexer."

I'm curious to know more about this. I was thinking about doing something similar using tmux (i.e. have one coding agent open up new ones in tmux and use 'send-keys' to control them). Is there any reason I might want to consider a different path? Something built on libghostty?

rahimnathwani··on The session you cannot take with you
How will Pi help? If I'm using pi with my ChatGPT subscription, that won't unlock access to full, unencrypted reasoning traces. If I switch a pi conversation from 5.6-sol to kimi k3, the 5.6-sol reasoning traces won't be visible to kimi k3. I think that's the main point of the article, no?
rahimnathwani··on Ask HN: How would you learn AI-assisted development from the ground up?
https://jiki.io/ was built for exactly this purpose, by the folks who made Exercism.
rahimnathwani··on Ask HN: How would you learn AI-assisted development from the ground up?
This is not a complete solution, but two books you should buy immediately:

Designing Data-Intensive Applications by Martin Kleppmann

AI Engineering by Chip Huyen

rahimnathwani··on LearnVector – Andrew Ng's AI company building one‑to‑one learning experiences
If you spend 30 mins/day studying math then that's $3.30 per hour.

If it improves your learning efficiency by just 10% then it's worth it.

(My son uses math academy 3-5 hours per week.)

rahimnathwani··on KOReader
Someone claims to have ported it to iOS but I haven't tried it and one of the open issues says that the release build doesn't work:

https://github.com/hezi/koreader-ios

rahimnathwani··on KOReader
IMO the most important one is more options for PDF rendering, often allowing you to read PDFs without squinting.
rahimnathwani··on Kids who own phones have poorer reading comprehension scores VS those who don't
No, they didn't control for confounders.

They did separate analysis based on grade level and whether English is the home language.

But they didn't control for parental education or socioeconomic status, both of which would likely be predictive of a child's reading comprehension and their chance of having a phone as a 9yo.

rahimnathwani··on Kids who own phones have poorer reading comprehension scores VS those who don't
Sure, a drive-by "did they control for confounders???" without naming any is low-effort, and it's reasonable to ask GP to be specific.

BUT you overshot in the other direction! "There's no such thing as controlling for confounders without blinded randomization" isn't fair. Sure, randomization is the cleanest identification strategy, but it's not the only one. Instrumental variables, regression discontinuity, difference-in-differences, sibling or twin fixed effects, and natural experiments are all used in economics, and they exist because randomization is often impossible or unethical, and you have to rely on natural experiments.

(Separately, a phone-ownership study of 9yo kids has an obvious potential for selection bias. So in a sense I'm sympathetic to GP asking about confounders without naming any.

rahimnathwani··on Simulate cassette tape audio profiles using FFmpeg
Back in the day, the plastic that wrapped blank tapes was printed with details about the tape inside, including a frequency response chart.

Of course it differed by brand.

It's also worth considering that a lot of tapes were recorded with Dolby B noise reduction, which boosted high frequencies on the recorded tape, to increase the signal to noise ratio. (You had to use Dolby B at the playback end, too, or it would sound too bright.)

rahimnathwani··on Harddrive Is Probably Full
If you're on Windows, MacOS or Linux, a great tool for interactively discovering and deleting large folders and files is dua:

https://github.com/Byron/dua-cli

If you're on MacOS and suspect a lot of space is taken by multiple copies of the same files, use fclones to turn them into APFS clones. (You can also use this on Linux, as ext4 supports hard links, but only if you're sure the duplicates won't diverge.)

← PreviousPage 5 of 34Next →